Based on improved deep convolutional neural network model pneumonia image classification.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computer-Aided Diagnosis

Background:

• Pneumonia is a leading infectious cause of death in children under five.
• Accurate diagnosis of pneumonia from X-ray images is crucial but can be subjective.
• Automated detection of lung abnormalities can improve diagnostic accuracy.

Purpose of the Study:

• To develop a deep learning model for automatic pneumonia diagnosis using X-ray images.
• To enhance diagnostic reliability in childhood pneumonia classification.
• To improve the accuracy and efficiency of pneumonia detection.

Main Methods:

• Utilized a deep learning approach combining the Xception neural network and Long Short-Term Memory (LSTM).
• Xception network extracted deep features, which were then processed by LSTM for detection and feature selection.
• Integrated Pearson's feature selection with a novel loss function to address category imbalance in training data.

Main Results:

• Achieved 96% accuracy, 99% area under the receiver operator characteristic curve, 98% precision, 91% recall, and 94% F1 score.
• Demonstrated superior performance compared to existing technical methods on available datasets.
• The model effectively assists doctors in classifying childhood pneumonia with high reliability.

Conclusions:

• The proposed Xception-LSTM model offers a reliable and accurate method for automated pneumonia detection in pediatric X-rays.
• This AI-driven approach has the potential to significantly improve the diagnosis of childhood pneumonia.
• The study highlights the effectiveness of combining advanced deep learning techniques for medical image analysis.